Sunday, October 30, 2005

The rankness of Google News

Interesting snippet from WebProNews, even if it is hidden inexplicably in a story about Digg getting funding (via benbarren):

Google News has garnered a following by gathering stories organized by topics, but an algorithm drives that process rather than users. In turn, Google News has been tweaked enough that only the most powerful news sources come to dominate topics, even when better stories from lesser sources may exist.

That's why a 100-word piece appearing on the Los Angeles Times web site will sit atop a topic in Google News while a lengthier and more useful 500-word report gets buried in the rankings for that topic.

If true, that would explain why GN is still in beta after three years. If you're presenting news results based on keyword searches, there are a number of criteria I can think of to produce a useful ranking, in the order I'd put them in:
  • number of times users clicked through on that story
  • presence of keyword in title
  • time since posted
  • source site
  • how early the keyword is first mentioned in article body
  • number of times keyword is mentioned in article body
  • length of article
  • number of inbound news links
  • number of inbound blog links
  • number of inbound Web links
  • how much the source site is paying me for high ranking
Oops, forget that last one. So aaaaanyway...


Post a Comment

Links to this post:

Create a Link

<< Home